[R] Positive Definite Matrix

[Kjetil Halvorsen]

see inline below.

On Tue, Feb 1, 2011 at 2:25 PM, Mike Marchywka <marchywka at hotmail.com> wrote:
>
>
>
>
>>Date: Tue, 1 Feb 2011 15:41:01 +0000
>>From: alex.smit4 at gmail.com
>>To: spencer.graves at structuremonitoring.com
>>CC: r-help at r-project.org; maechler at stat.math.ethz.ch
>>Subject: Re: [R] Positive Definite Matrix
>
>
>>Hello thank you for so much input. I am afraid that I am fairly new to this
> and some of the material is above my understanding for the time being and I
> am not looking at anything very complex as you will see from the program I
> will include.
> I have been talking to a fellow colleague I am working with and have come
> down to this.
> Your input has been invaluable to me learning more about R thank you!
> 1.As an example, I do not wish to find a near pd matrix but I want to test
> wether it is pd or not so I know if I want it or not,as you will see on the
> program I want two outputs depending on if it is a pd and symmetric. So I
> have written this for one symmetric matrix and it works and one for a non
> symmetric one slightly changed but it doesnt work,if you could point out why
> this is going wrong that would be helpful. It cant compare the complex
> eigenvalues.
> [...]
>
>>2. My colleague said why dont we use the Sylvester's Criterion and test the
> leading minors. Would something like that work?

It would work, but it would be more work than just testing the eigenvalues!

kjetil

>
>
> This thread seems to have dragged on for a while, what are you trying to do?
> Presumably this is something other than an exercize in R.
> What interpretation do you have for your eignenvalue spectrum or matrix element
> ? Problems here could point to various features or pathologies or
> opportunities at various points in the analysis.
>
>
>
>
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>